gemini-cli-extensions/cloud-sql-sqlserver
claude-codeOfficialSkills for Cloud SQL for SQL Server
Cloud SQL for SQL Server agent skills is an open-source extension by Google, part of its Gemini CLI Extensions family, that lets AI coding agents provision, query, and manage GCP SQL Server databases through natural language, no raw T-SQL required.
Our Review
Maintained by Google as part of the official Gemini CLI Extensions family, this Apache-2.0 project packages four task-scoped skill groups that let an AI agent operate a Cloud SQL for SQL Server instance from the same chat where you write code. Instead of memorizing gcloud flags and T-SQL, you describe the task and the agent runs it.
What this extension adds to your AI agent:
- •Admin skills provision new Cloud SQL for SQL Server instances, create databases and users, clone environments, and track long-running operations from a prompt.
- •Data skills explore schemas, run SQL queries against your tables, and read system performance metrics with PromQL, all described in plain English.
- •Lifecycle skills create backups, restore from existing backups, and clone instances for testing or migration without leaving the agent.
- •Monitor skills troubleshoot slow queries and analyze system-level PromQL metrics to find the cause of a performance problem.
- •Code generation ask the agent to generate data classes and snippets, such as a Python dataclass mapped to a real table schema.
- •MCP Toolbox backend the skills call Google's MCP Toolbox under the hood, so database access goes through a maintained, GCP-native layer.
Compatibility:
The skills install into three AI agents: Antigravity (CLI v1.6.0+ or Antigravity 2.0), Claude Code (v2.1.94+, via its plugin marketplace), and Codex (v0.117.0+). You can also register them with the open npx skills tool. Each agent reads the same skill folders, so the behavior is consistent across hosts.
Setup needs a Google Cloud project with the Cloud SQL Admin API enabled, Application Default Credentials in your environment, and IAM roles for Cloud SQL Client, Viewer, and Admin. Connection details (project, region, instance, database, user, password) are passed through CLOUD_SQL_MSSQL_* environment variables.
Limitations:
This is beta software (pre-v1.0), so the project warns that breaking changes can land before the first stable release. It only works with Cloud SQL for SQL Server on Google Cloud, not self-managed SQL Server, Azure SQL, or AWS RDS, and it requires GCP credentials plus several IAM roles, which puts it out of reach for anyone without a Google Cloud project. Private-IP instances require the agent to run inside the same VPC network. The repo is young (7 stars as of June 2026), so community examples are still thin compared with mature database tooling.
Our Verdict
The cloud-sql-sqlserver extension is worth installing if you already run Cloud SQL for SQL Server on Google Cloud and want your AI agent to handle routine database work. It removes the context switch between writing code and managing the database, which is the daily friction this extension is built to solve.
For Vibe Builders, the appeal is plain-English control of a real production database. You can ask for a new instance, a backup, or a query result without learning gcloud or T-SQL, and the four skill groups cover the tasks a small team hits most often.
For Developers, the value is that Google maintains it and it routes through MCP Toolbox, so you get a GCP-native data layer instead of a hand-rolled MCP server. It slots into Claude Code, Codex, or Antigravity with the same skill folders, so your whole team can standardize on one toolset.
Skip it if you do not use Cloud SQL for SQL Server on Google Cloud, if you need a single tool that spans Azure or AWS databases, or if you cannot tolerate beta-stage breaking changes in a production workflow. For a fully managed vector database to pair with your app, Pinecone is a better fit, and for the general AI coding workflow itself, Cursor and Claude Code remain the agents you would install these skills into. At a beta 7.5 out of 10, this is a focused, official extension that earns its place for GCP SQL Server teams.
Frequently Asked Questions
What is the Cloud SQL for SQL Server agent skills extension?
It is an open-source extension by Google, published in its Gemini CLI Extensions family, that gives AI coding agents a set of skills for managing Cloud SQL for SQL Server databases on Google Cloud. Released under the Apache-2.0 license and active in 2026, it lets you provision instances, run queries, explore schemas, and troubleshoot performance using natural-language prompts instead of raw T-SQL or gcloud commands.
Is the Cloud SQL for SQL Server extension free and open-source?
Yes. The project is open-source under the Apache-2.0 license and free to install from its GitHub repository. You pay only for the underlying Google Cloud resources you create, such as Cloud SQL instances and storage. There is no separate license fee for the agent skills themselves, and the source code is fully available for review or modification.
Which AI agents work with these Cloud SQL skills?
The skills install into three AI agents: Antigravity (CLI v1.6.0 or higher, or Antigravity 2.0), Claude Code (v2.1.94 or higher, through its plugin marketplace), and Codex (v0.117.0 or higher). You can also register them with the open npx skills tool. All three agents read the same skill folders, so behavior stays consistent across hosts.
What can the Cloud SQL for SQL Server skills actually do?
They cover four task groups: Admin for provisioning instances, databases, and users; Data for exploring schemas and running SQL queries; Lifecycle for creating backups, restoring, and cloning instances; and Monitor for troubleshooting slow queries with PromQL metrics. The extension can also generate code, such as a Python dataclass that maps to a real table schema, directly from your prompts.
What do I need to install and use the extension in 2026?
You need a Google Cloud project with the Cloud SQL Admin API enabled, Application Default Credentials set up in your environment, and three IAM roles: Cloud SQL Client, Viewer, and Admin. Connection details are supplied through CLOUD_SQL_MSSQL environment variables, and as of 2026 you add the marketplace inside your agent and install the plugin. Private-IP instances require running the agent in the same VPC network.
What is cloud-sql-sqlserver?
Cloud SQL for SQL Server agent skills is an open-source extension by Google, part of its Gemini CLI Extensions family, that lets AI coding agents provision, query, and manage GCP SQL Server databases through natural language, no raw T-SQL required.
How do I install cloud-sql-sqlserver?
Visit the GitHub repository at https://github.com/gemini-cli-extensions/cloud-sql-sqlserver for installation instructions.
What license does cloud-sql-sqlserver use?
cloud-sql-sqlserver uses the Apache-2.0 license.
What are alternatives to cloud-sql-sqlserver?
Explore related tools and alternatives on My AI Guide.
Open source: source code publicly visible
Anyone can inspect exactly what this repo does on GitHub before using it. Apache-2.0 licensed.
Reviewed by My AI Guide for relevance, quality, and active maintenance before listing.
Install in Claude Code:
/install cloud-sql-sqlserver